Grantmatch AI

GrantMatch

Inspiration

Right now, non-profit organizations (NPOs) are drowning in information overload. The OurSG Grants portal lists over 100 grants, but finding the right one requires exhausting manual reviews and basic keyword searches that miss semantically relevant opportunities. Staff are stuck using manual spreadsheets to track deadlines and eligibility, often spending hours on applications before realizing they aren't even a good fit.

What it does

GrantMatch AI is a web-based platform designed to simplify grant discovery through the power of Natural Language Processing (NLP). Instead of fighting with filters, NPOs can simply describe their project needs in plain English

Key Features

1) Semantic Search: Our AI understands natural language queries to find grants that actually match an organization's mission.

2) Personalized Matching: We layer manual filters on top of AI results, injecting the organization's specific profile into every search for high-relevance results.

3) Instant Analysis: With one click, the platform generates a "Why You’re a Good Fit" write-up, tailoring the grant’s requirements to the organization’s specific strengths to jumpstart the proposal process.

4) Proactive Discovery: Users receive personalized email alerts and deadline reminders, ensuring they never miss a funding opportunity again.

Challenges we faced

1) Real-time Data Sync: The web app does not currently have the capability to update its database from OurGrantsSG in real-time.

2) API Constraints: The "LLM brain" utilizes the Groq API, which is limited in capability on the free tier (70B Llama-3.3). Multiple API keys are currently linked in parallel to circumvent rate limiting.

3) Notification Optimization: The email notification feature has not been optimized to work with the LLM for further personalization. It currently uses manual parameters to filter grants.

What is next for Grantmatch AI

1) Platform Integration: As usage gains traction, the team hopes to streamline the platform with the OurGrantsSG dev team to link seamless access to grant updates.

2) Model Overhaul: There are plans to either train a custom ML model with NLP reasoning or combine existing technologies with more powerful reasoning AI.

3) AI-Filtered Emails: The team plans to integrate the LLM to filter emails for each user, though this may be difficult as the user base grows.